|Publication number||US6949951 B1|
|Application number||US 10/869,777|
|Publication date||Sep 27, 2005|
|Filing date||Jun 15, 2004|
|Priority date||Jan 30, 2003|
|Also published as||US6768335|
|Publication number||10869777, 869777, US 6949951 B1, US 6949951B1, US-B1-6949951, US6949951 B1, US6949951B1|
|Inventors||Steven P. Young, Michael J. Hart, Venu M. Kondapalli, Martin L. Voogel|
|Original Assignee||Xilinx, Inc.|
|Export Citation||BiBTeX, EndNote, RefMan|
|Patent Citations (16), Non-Patent Citations (3), Referenced by (18), Classifications (18), Legal Events (3)|
|External Links: USPTO, USPTO Assignment, Espacenet|
The invention relates to programmable logic devices (PLDs). More particularly, the invention relates to an interconnect multiplexer for a PLD that includes transistors having more than one oxide thickness.
Programmable logic devices (PLDs) are a well-known type of digital integrated circuit that can be programmed to perform specified logic functions. One type of PLD, the field programmable gate array (FPGA), typically includes an array of configurable logic blocks (CLBs) and programmable input/output blocks (IOBs). The CLBs and IOBs are interconnected by a programmable interconnect structure.
More advanced FPGAs can include more than one type of logic block in the array. For example, the Xilinx Virtex-II Pro™ FPGA includes blocks of Random Access Memory (RAM), blocks implementing multiplier functions, and embedded processor blocks. (The Xilinx Virtex-II Pro FPGA is described in detail in pages 19–71 of the “Virtex-II Pro Platform FPGA Handbook”, published October, 2002, available from Xilinx, Inc., 2100 Logic Drive, San Jose, Calif. 95124, which pages are incorporated herein by reference.)
The CLBs, IOBs, interconnect, and other logic blocks are typically programmed by loading a stream of configuration data (bitstream) into internal configuration memory cells that define how the CLBs, IOBs, and interconnect are configured. The configuration data can be read from memory (e.g., an external PROM) or written into the FPGA by an external device. The collective states of the individual memory cells then determine the function of the FPGA.
The programmable interconnect structure typically includes a large number of interconnect lines of various lengths interconnected by programmable interconnect points (PIPS). These PIPs are controlled by configuration memory cells. Thus, the values stored in the PIP configuration memory cells determine the “routing”, i.e., the interconnections between various points in the FPGA. PIPs can also be used to select a signal from one of several interconnect lines and to pass the selected signal to a CLB, for example, or to select one of several output signals from a CLB to pass to an interconnect line.
A PIP can be implemented in various ways. For example, a set of PIPS can be implemented as a multiplexer structure that selects one of several interconnect lines and drives a signal on the selected interconnect line onto a destination interconnect line. One such set of PIPs is shown in
As shown in
The structure shown in
When a high value is passed through an N-channel transistor, the voltage is reduced by Vt, the threshold voltage level of the N-channel transistor. (For purposes of simplifying the present discussion, other well-known transistor effects such as the body effect are not described.) For example,
This well-known limitation presents no problem when VDD=5 V, as described above. When processes supporting this operating voltage are used, the N-channel threshold voltage is less than one volt (1 V). Thus, the voltage at internal node INT is still more than four volts (4 V), which is quite sufficient to turn on the N-channel pulldown N1, thus turning on P-channel pullup P2 in buffer 201 (see
For example, if node INT does not rise sufficiently high, the N-channel pulldown in buffer 201 might not be completely on, and the P-channel pullup P2 might not be sufficiently on to quickly turn off pullup P1. Thus, the pullup (e.g., pullup P1) “fights” the pulldown (e.g., pulldown N1), and slows down the switching of the output node. For example, referring again to
As described, there is a voltage drop when a high voltage is passed across an N-channel transistor. However, there is no corresponding voltage drop of a high voltage across a P-channel transistor. Therefore, one solution to this problem is to replace the N-channel transistor with paired N- and P-channel transistors, with the P-channel transistors being driven by the complementary input signal. However, this solution can double the size of the multiplexer by doubling the number of pass transistors. Further, both the true and complement signals must be provided to the paired transistors, which increases the die area. Larger buffers must also be provided to handle the increased load of driving both the N- and P-channel transistors.
Another way to accommodate lower VDD levels is to drive the gates of the N-channel transistors with a voltage level VGG higher than VDD, as shown in
However, FPGA manufacturers are not only reducing operating voltage levels on their devices, but are also reducing minimum feature sizes to lower costs. Sometimes, each new product generation uses a lower operating voltage and has reduced feature sizes. Reduced feature sizes mean thinner oxide and shorter gate lengths. Therefore, specified values of VGG are being rapidly reduced. For example, Table 1 shows the relationships between various exemplary values of VDD (the operating voltage), VGG (the maximum allowable gate voltage), Vt (the N-channel threshold voltage), and VINT (the voltage at node INT). Also included are typical gate lengths for each manufacturing process. All of the values shown in Table 1 are merely exemplary, as they vary between processes and manufacturers. VDD values are specified as nominal operating voltages.
VINT = VGG − Vt
Note that as VGG approaches twice the threshold voltage (VGG=2Vt), the voltage at node INT becomes insufficient to adequately turn on pulldown N1 (see
The invention provides a multiplexer structure that can be used, for example, in the programmable interconnect points (PIPs) of a programmable logic device (PLD). The structure includes transistors of varying oxide thickness, operating voltage, and, in some embodiments, gate length.
The multiplexer includes a plurality of pass transistors passing a selected one of several input values to an internal node, which drives a buffer providing the multiplexer output signal. The pass transistors can be controlled, for example, by values stored in memory cells of a PLD. The pass transistors have a first oxide thickness and are controlled by a value having a first operating voltage. The buffer includes transistors having a second oxide thickness thinner than the first oxide thickness, and operates at a second operating voltage lower than the first operating voltage. In some embodiments, the second operating voltage is the operating voltage used as a supply voltage by a majority of the transistors in the PLD. Where memory cells are used to control the pass transistors, the memory cells operate at the first operating voltage and include transistors having the first oxide thickness.
In some embodiments, the gate length also varies between the pass transistors, buffer transistors, and memory cell transistors. To support the higher operating voltage with acceptable leakage, the memory cells have the longest gate length. The buffer transistors, on the other hand, operate at the lower voltage (with thinner oxide) and can have the shortest gate length without generating unacceptable leakage current. The pass transistors are gated by the higher voltage and thus require the thicker oxide. However, the drain-to-source voltage for the pass transistors is less than the higher operating voltage, so the gate length of the pass transistors can be somewhere between the longest and shortest gate lengths without generating unacceptable leakage current.
In other embodiments, the pass transistors and memory cell transistors have a first and longer gate length, while the buffer transistors have a second and shorter gate length. In yet other embodiments, the gate lengths are the same for the pass transistors, buffer transistors, and memory cell transistors.
According to some embodiments, a multiplexer structure in an integrated circuit (IC) includes a plurality of pass transistors coupled between a plurality of input terminals and an internal node, and a buffer coupled between the internal node and an output terminal. Each of the pass transistors has a gate terminal coupled to provide to the pass transistor a selected one of a ground signal and a power high signal at a first voltage level. Each of the pass transistors has a first oxide thickness. The buffer is coupled to an operating voltage at a second voltage level less than the first voltage level. The buffer comprises buffer transistors having a second oxide thickness, where the second oxide thickness is less than the first oxide thickness. In some embodiments, the multiplexer structure also includes a plurality of memory cells controlling the pass transistors, operating at the first voltage level, and having the first oxide thickness.
In some embodiments, the IC is a PLD such as a field programmable gate array (FPGA). In some such embodiments, the multiplexer structure is used to implement programmable interconnect points (PIPs) in the FPGA. In other embodiments, the IC is a non-programmable integrated circuit.
According to some embodiments, a circuit comprises a pass transistor coupled between an input terminal and an internal node and a buffer coupled between the internal node and an output terminal. The pass transistor has a first oxide thickness and a gate terminal coupled to provide to the pass transistor a selected one of a ground signal and a power high signal at a first voltage level. The buffer comprises buffer transistors having a second oxide thickness less than the first oxide thickness, and is coupled to an operating voltage at a second voltage level less than the first voltage level. In some embodiments, the circuit also includes a memory cell controlling the pass transistor, operating at the first voltage level, and having the first oxide thickness.
According to some embodiments, an interconnect structure for a PLD includes a plurality of interconnect lines and a first plurality of PIPs coupled between the interconnect lines. Each of the first plurality of PIPs comprises a plurality of input terminals coupled to a subset of the interconnect lines and an output terminal coupled to one of the interconnect lines. Each set of PIPs is implemented in a manner substantially similar to the multiplexer structure as described above.
According to some embodiments, an FPGA comprises a plurality of configurable logic blocks, a plurality of interconnect lines, and a plurality of PIPs coupled between the interconnect lines and the configurable logic blocks. Each set of PIPs is implemented in a manner substantially similar to the multiplexer structure as described above.
The present invention is illustrated by way of example, and not by way of limitation, in the following figures.
The present invention is believed to be applicable to a variety of integrated circuits, including but not limited to programmable logic devices (PLDs). The present invention has been found to be particularly applicable and beneficial when applied to field programmable gate arrays (FPGAs), and in particular to programmable interconnect points (PIPS) in FPGAS. While the present invention is not so limited, an appreciation of the present invention is presented by way of specific examples, which in this instance are directed to such FPGA PIPs.
In the following description, numerous specific details are set forth to provide a more thorough understanding of the present invention. However, it will be apparent to one skilled in the art that the present invention can be practiced without these specific details.
In the pictured embodiment, buffer 401 is implemented as a CMOS inverter. However, many other buffer implementations can be used. For example, buffer 101 x of
In the pictured embodiment, memory cell C is implemented as two cross-coupled inverters (P-channel transistors P5, P6 and N-channel transistors N5, N6). However, many other memory cell implementations can be used, or the gate terminal of pass transistor G can be driven by a circuit other than a memory cell. Whether a memory cell or some other circuit is used, the circuit provides to the gate of the pass transistor either the ground value or a voltage level VGG higher than the operating voltage VDD of the buffer.
Both the pass transistor G and the memory cell C are manufactured using a thicker oxide layer than the buffer transistors (see area 410 in
As explained above in connection with
(= VGG − Vt (thick))
Comparing Table 2 to the last line of Table 1, it can be seen that by using thicker oxide for the pass transistor and raising the gate voltage to 1.5 V, the voltage at internal node INT has increased from 0.8 V to 0.95 V, thereby enhancing the performance of the circuit.
For a given source/drain voltage and oxide thickness, an IC manufacturer typically specifies a minimum gate length. The specified minimum gate is generally longer for gates with a thicker oxide. Thus, when a higher operating voltage is used, in addition to using a thicker oxide it is generally necessary to increase the gate length of any minimum-sized transistors involved. Further, when a higher voltage (e.g., VGG instead of VDD) is applied across a transistor from source to drain, leakage current increases. This leakage current can be reduced by making the transistors longer, i.e., increasing the gate length. Note that in memory cell C of
However, note also that the maximum drain-source voltage across pass transistor G is not VGG, but VDD. (The maximum voltage in this case occurs when node IN is switching levels in either direction.) Therefore, the gate length of pass transistor G need not necessarily be as long as those of memory cell transistors P5, N5, P6, and N6 to have an acceptable leakage current. Thus, in some embodiments the buffer transistors have the shortest gate length, the memory cell transistors have the longest gate length, and the pass transistor has a gate length somewhere between the two.
In some embodiments, the gate lengths of the N-channel transistors and the P-channel transistors are not the same. For example, in some embodiments leakage current is higher for N-channel devices. Therefore, in these embodiments it can be desirable to use a somewhat longer gate length for the N-channel transistors than for corresponding P-channel transistors, to reduce leakage.
Buffer 501 x is one embodiment of buffer 501 that can be used to implement, for example, a set of FPGA PIPs. Buffer 501 x is similar to buffer 101 x of
Multiplexer structure 500 includes eight pass transistors G0–G7. However, any number of pass transistors can be used, subject to standard design considerations such as layout area and capacitance on node INT. Further, one or more levels of pass transistors can be used. Such design considerations are well known in the IC design art. Pass transistors G0–G7 and memory cells C0–C7 are included in area 510, where a thick oxide is used to implement the transistors.
Pass transistors G0–G7 are controlled by memory cells C0–C7, respectively. Memory cells C0–C7 can be implemented, for example, using two cross-coupled inverters as shown in inset Cx. Some embodiments include the configuration memory cell from the Xilinx Virtex-II Pro FPGA, with thick oxide being used to implement the cross-coupled gates and the pass transistors used to write new values to the memory cell. In a PLD, memory cells can include other well-known logic such as configuration logic, initialization logic, and/or readback logic, which are not shown in
Multiplexer structure 500 can be used, for example, to implement a set of FPGA PIPs. In these embodiments, at any given time only one of memory cells C0–C7 is programmed to enable the associated pass transistor G0–G7.
In some embodiments, the operating voltage VDD of the buffer is the same as the operating voltage used for the majority of transistors in the FPGA.
Area 720 includes buffers 702, 703, which can be implemented in various ways, but in which the transistors driven by internal nodes INTB and INTA have a relatively thin oxide and short gate length. In some embodiments, the buffers are implemented in a fashion similar to buffer 601 x (see
Area 711 includes memory cells D0–D13, which can also be implemented in various ways, but in which the transistors driving the gate terminals of the pass transistors use a relatively thick oxide and have a relatively long gate length. Memory cells D0–D13 operate at a second and higher voltage level than buffers 702, 703.
Area 712 includes pass transistors E0–E11, F0–F3, Q0–Q11, and P0–P3. The pass transistors are implemented using the relatively thick oxide used in area 711, and have gate lengths shorter than those of areas 711 but longer than those of area 720. The pass transistors operate at (i.e., their gate terminals are driven by) the second and higher voltage level.
Table 3 shows the operating voltage, oxide thickness, maximum source/drain voltage, and gate length used in one embodiment for each area of
In some PLD embodiments, the pass transistors are not controlled directly by the memory cells. Instead, fewer memory cells are used, and the values stored in the memory cells are decoded to provide the gate control signals to the pass transistors.
In some embodiments, the multiplexer structures of the invention are included in an IC manufactured using three different oxide thicknesses. The thinnest oxide is used for the buffers in the multiplexer structures (and for other circuits in the device as well). The medium-thickness oxide is used for the pass transistors and memory cells in the multiplexer structures. The thickest oxide is used in the input/output (I/O) structures.
It is well known to use a thicker oxide in the I/O structures of integrated circuits. It is not uncommon for input signals provided to an IC to have a high voltage value that is greater than the core operating voltage of the device. For example, ICs with a core operating voltage of 1.5 V and a VGG value of 1.6 volts might be required to interface with other ICs that provide and receive signals at 2.5 volts. To support these higher voltages, I/O structures are commonly manufactured with a much thicker oxide than the internal structures of the IC.
However, it is not generally desirable to use this much thicker oxide in internal structures such as the multiplexer structures provided herein. The much thicker oxide requires very large transistors and generally results in much worse performance than the medium-thickness oxide described herein.
Those having skill in the relevant arts of the invention will now perceive various modifications and additions that can be made as a result of the disclosure herein. For example, the above text describes the structures of the invention in the context of ICs such as programmable logic devices (PLDs) and field programmable gate arrays (FPGAs). However, the structures of the invention can also be implemented in other integrated circuits, including non-programmable circuits. Additionally, in PLDs and FPGAs the structures are described in the context of programmable interconnect points (PIPs) between the interconnect lines and configurable logic blocks (CLBs) of the device. However, the structures of the invention can also be included in other circuits and for other purposes.
Further, buffers, memory cells, pass transistors, P-channel transistors, N-channel transistors, pullups, pulldowns, multiplexers, multiplexer structures, interconnect structures, PIPS, configurable logic blocks, FPGAs, PLDS, ICs, and other components other than those described herein can be used to implement the invention. Active-high signals can be replaced with active-low signals by making straightforward alterations to the circuitry, such as are well known in the art of circuit design. Logical circuits can be replaced by their logical equivalents by appropriately inverting input and output signals, as is also well known.
Moreover, some components are shown directly connected to one another while others are shown connected via intermediate components. In each instance the method of interconnection establishes some desired electrical communication between two or more circuit nodes. Such communication can often be accomplished using a number of circuit configurations, as will be understood by those of skill in the art.
Accordingly, all such modifications and additions are deemed to be within the scope of the invention, which is to be limited only by the appended claims and their equivalents.
|Cited Patent||Filing date||Publication date||Applicant||Title|
|US4912348 *||Dec 9, 1988||Mar 27, 1990||Idaho Research Foundation||Method for designing pass transistor asynchronous sequential circuits|
|US5059828 *||Mar 23, 1990||Oct 22, 1991||Oki Electric Industry Co., Ltd.||Programmable logic array circuit having a gate to control an output condition state of a latch thereof|
|US5744995||Apr 17, 1996||Apr 28, 1998||Xilinx, Inc.||Six-input multiplexer wtih two gate levels and three memory cells|
|US5801551||Aug 1, 1996||Sep 1, 1998||Advanced Micro Devices, Inc.||Depletion mode pass gates with controlling decoder and negative power supply for a programmable logic device|
|US5808483||Sep 20, 1996||Sep 15, 1998||Kawasaki Steel Corporation||Logic circuit utilizing pass transistors and logic gate|
|US5955912||Oct 24, 1996||Sep 21, 1999||Texas Instruments Incorporated||Multiplexer circuits|
|US6049227||Nov 5, 1998||Apr 11, 2000||Xilinx, Inc.||FPGA with a plurality of I/O voltage levels|
|US6087886 *||Apr 8, 1999||Jul 11, 2000||Texas Instruments Incorporated||Hybrid dual threshold transistor multiplexer|
|US6278290 *||Aug 13, 1999||Aug 21, 2001||Xilinx, Inc.||Method and circuit for operating programmable logic devices during power-up and stand-by modes|
|US6373291||Sep 25, 2000||Apr 16, 2002||Kabushiki Kaisha Toshiba||Pass transistor logic circuit for reducing power consumption|
|US6380765||Aug 29, 2000||Apr 30, 2002||Micron Technology, Inc.||Double pass transistor logic with vertical gate transistors|
|US6392467||Jul 10, 2000||May 21, 2002||Toshiba Corporation||Semiconductor integrated circuit|
|US6448809||Aug 7, 2001||Sep 10, 2002||Xilinx, Inc.||FPGA with a plurality of input reference voltage levels|
|US6667635||Sep 10, 2002||Dec 23, 2003||Xilinx, Inc.||FPGA lookup table with transmission gate structure for reliable low-voltage operation|
|US6768338||Jan 30, 2003||Jul 27, 2004||Xilinx, Inc.||PLD lookup table including transistors of more than one oxide thickness|
|US6798270 *||Jul 11, 2003||Sep 28, 2004||Xilinx, Inc.||Pass gate multiplexer circuit with reduced susceptibility to single event upsets|
|1||U.S. Appl. No. 10/354,587, filed Jan. 30, 2003, Young et al.|
|2||U.S. Appl. No. 10/869,139, filed Jun. 15, 2004, Young et al.|
|3||Xilinx, Inc.; "Virtex-II Pro Platform FPGA Handbook"; published Oct. 14, 2002; available from Xilinx, Inc., 2100 Logic Drive, San Jose, California 95124; pp. 19-71.|
|Citing Patent||Filing date||Publication date||Applicant||Title|
|US7221192 *||Jan 13, 2005||May 22, 2007||Advanced Micro Devices, Inc.||Voltage access circuit configured for outputting a selected analog voltage signal for testing external to an integrated circuit|
|US7245152 *||May 2, 2005||Jul 17, 2007||Atmel Corporation||Voltage-level shifter|
|US7423452||Jul 27, 2007||Sep 9, 2008||Xilinx, Inc.||Integrated circuit including a multiplexer circuit|
|US7463067 *||Oct 2, 2006||Dec 9, 2008||Stmicroelectronics S.R.L.||Switch block for FPGA architectures|
|US7471104||Mar 31, 2008||Dec 30, 2008||Xilinx, Inc.||Lookup table with relatively balanced delays|
|US7545177||Mar 20, 2007||Jun 9, 2009||Xilinx, Inc.||Method and apparatus for leakage current reduction|
|US7683674 *||Jun 6, 2007||Mar 23, 2010||Stmicroelectronics, S.R.L.||T-switch buffer, in particular for FPGA architectures|
|US7804321 *||Sep 5, 2008||Sep 28, 2010||Actel Corporation||Circuits and methods for testing FPGA routing switches|
|US7893712 *||Sep 10, 2009||Feb 22, 2011||Xilinx, Inc.||Integrated circuit with a selectable interconnect circuit for low power or high performance operation|
|US7919977 *||Aug 20, 2010||Apr 5, 2011||Actel Corporation||Circuits and methods for testing FPGA routing switches|
|US9000809 *||May 30, 2012||Apr 7, 2015||Infineon Technologies Austria Ag||Method and device for sampling an input signal|
|US20060279332 *||May 2, 2005||Dec 14, 2006||Wich Mathew T||Voltage-level shifter|
|US20070085563 *||Oct 2, 2006||Apr 19, 2007||Stmicroelectronics S.R.L.||Switch block for fpga architectures|
|US20070279088 *||Jun 6, 2007||Dec 6, 2007||Stmicroelectronics S.R.L.||T-switch buffer, in particular for FPGA architectures|
|US20100060311 *||Sep 5, 2008||Mar 11, 2010||Actel Corporation||Circuits and methods for testing fpga routing switches|
|US20100315118 *||Aug 20, 2010||Dec 16, 2010||Actel Corporation||Circuits and methods for testing fpga routing switches|
|WO2006119276A2 *||May 2, 2006||Nov 9, 2006||Atmel Corporation||Voltage-level shifter|
|WO2006119276A3 *||May 2, 2006||Mar 29, 2007||Atmel Corp||Voltage-level shifter|
|U.S. Classification||326/37, 326/38, 326/39, 327/408, 327/407, 326/113|
|International Classification||G06F7/38, H03K17/00, H03K19/177, H03K17/693|
|Cooperative Classification||H03K19/17736, H03K17/693, H03K19/1778, H03K17/005|
|European Classification||H03K19/177J, H03K19/177F, H03K17/00M2, H03K17/693|
|Jun 15, 2004||AS||Assignment|
Owner name: XILINX, INC., CALIFORNIA
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNORS:YOUNG, STEVEN P.;KONDAPALLI, VENU M.;HART, MICHAEL J.;AND OTHERS;REEL/FRAME:015492/0831;SIGNING DATES FROM 20030122 TO 20030127
|Mar 23, 2009||FPAY||Fee payment|
Year of fee payment: 4
|Oct 2, 2012||FPAY||Fee payment|
Year of fee payment: 8